Development of Hollow Fiber Carbon Membranes for CO2 Separation

Abstract

CO2 capture from flue gases by membrane technology in post combustion power plants could be used for the reducing of CO2 emissions. Previous work has demonstrated that the carbon membrane can achieve a high separation performance with respect to high CO2 permeability and selectivity over the other gases, such as N2 and O2. The focus of the current work was to find a low-cost precursor and develop a simple process for the preparation of high performance hollow fiber carbon membranes (HFCMs) for CO2 separation.

The cellulose acetate (CA) hollow fibers were spun from a dope solution of CA / Polyvinylpyrrolidone (PVP) / N-methyl-2-pyrrolidone (NMP) (22.5 % / 5 % / 72.5 %) using an optimal spinning condition: bore fluid (water + NMP (85 %)), air gap (25 mm), bore flow rate (40 % of dope flow rate) and temperature of quench bath (50 °C). The cellulosic hollow fibers, regenerated from the spun CA fibers by deacetylation, were used as the precursors for preparation of HFCMs. The experimental results indicated that the precursors would influence significantly the separation performances of the prepared HFCMs. Therefore, the deacetylation process needed to be optimized, and the optimal deacetylation condition was found to be: soaking CA hollow fibers in a 10 % glycerol solution for 24 h, and then treated by immersion in a 0.075 M NaOH (96 % ethanol) solution for 2 h.

The carbonization parameters were also found to affect the separation properties of the HFCMs significantly. The carbonization condition was optimized based on an orthogonal experimental design method and statistical analysis. The optimal carbonization procedure was found to be: CO2 as purge gas, a final temperature of 823K, a heating rate of 4K/min and a final soak time of 2h (CO2-823K-4K/min-2h), and the importance of the investigated carbonization parameters was sorted out with respect to their influence on carbon membrane separation performances. The order of importance for the carbonization parameters was found to be: purge gas > final temperature > heating rate > final soak time. It was hence concluded that the purge gas was the most important parameter affecting the final carbon membrane performance, and CO2 was found to be the most effective purge gas for preparation of the high performance cellulosic derived carbon membranes.

A symmetric structure for the prepared HFCMs with a typical thickness of 25 um was identified from the scanning electronic microscopy (SEM) images, and a great shrinkage compared to the precursor could be seen. The Fourier Transform Infrared (FTIR) spectra showed the decomposition and break down of the chemical groups in precursors in various carbonization environments, leading to the release of volatile gases. A typical d-spacing of the carbon membranes was found to be 4 A from the X-ray diffraction (XRD) characterization. CO2 and N2 adsorption equilibrium isotherms were obtained by the gravimetric sorption measurements. A higher adsorption affinity of CO2, obtained by fitting the experimental data using Langmuir-Freundlich model, indicated that CO2 is more adsorbable than N2. Two type of hollow fiber carbon membrane (HFCM-A and HFCM-B) has been prepared in the current work. The micropore volume and average pore size for the carbon membrane HFCM-B are around 0.17 cm3/g and 5.6 A, respectively, which are slightly larger than that of the HFCM-A (0.15 cm3/g and 5.2 A). The kinetic rate constants were also determined from the CO2 kinetic adsorption experiments, and the higher kinetic rate constant of HFCM-B indicates the more open structure.

Gas permeation tests were conducted with single gases (H2, CO2, O2, N2 and CH4) as well as gas mixtures. The single gas permeation tests confirmed that the permeability values decreased with increasing kinetic diameter of the gas molecules, which indicated that the molecular sieving mechanism was dominating in the carbon membrane separation processes. The results also showed that the kinetic diameter had a larger effect than the Lennard-Jones well depth, which indicated that the diffusion was dominated by a molecular sieving process and that the sorption had relatively little influence. The gas permeability increased with temperature due to the activated transport process for the molecular sieve mechanism. The gas molecules with larger activation energy (e.g. CH4 and N2) were affected by the temperature more significantly, comparing to that with lower activation energy (e.g. CO2). The strongly adsorbed CO2 showed a more significant decrease of permeability with pressure compared to N2, reflecting a stronger concentration dependence for the diffusion coefficient of CO2. The gas permeability decreased with the presence of water vapor which might be caused by the pore blocking. The aging test results indicated that the permeability of carbon membrane decreased over time when exposed to air, and needed to be regenerated. The gas mixture measurements showed that the significant effects of the operating parameters, especially the feed pressure, on the membrane performance based on the fractional factorial design method and statistical analysis. Therefore, the operating conditions need to be optimized for the specific applications.

The single stage membrane processes for CO2 capture from flue gases with feed compression, permeate evacuation, and their combination were investigated using Aspen HYSYS simulation tool integrated with an in-house membrane simulation model called ChemBrane. The simulation results indicated that the single stage membrane process could not achieve high CO2 purity and CO2 recovery simultaneously using these HFCMs. The plotted characteristic diagrams could be easily used to identify the required operating conditions and membrane areas to accomplish specific targets for a given separation process. A two stage membrane system was also designed for the evaluation of process feasibility, and the simulation results indicated that a CO2 purity of 90 % and a recovery of 80 % could be achieved by optimizing of the process conditions. Although the cost of carbon membranes is still unknown, the membrane/module cost could be greatly reduced by further improving the membrane separation performance (especially increasing the gas permeance by reducing the wall thickness of HFCMs) and simplifying the membrane production process. The capital cost estimation for two-stage cascade membrane process indicated that the potential application for carbon membrane technique could be promising compared to chemical absorption.